My PC has been unstable for much of this year. I previously replaced an SSD after tests and Event Viewer reported bad blocks, but the system still randomly freezes. During gaming, the freeze often begins while the game is saving: the save indicator runs indefinitely, File Explorer stops responding, and video playback may continue briefly with audio still working before the entire system locks up and requires a hard reboot. There is no BSOD, minidump, or useful Event Viewer entry.
Reseating the graphics card seemed to help for a few days, but the problem returned. CPU stress testing appears stable, while GPU stress testing sometimes causes the issue, although it is not always reproducible. The system never freezes during light use such as watching videos, and temperatures look normal.
The hardware is a Ryzen 9 5900X, ASUS ROG Strix X570-E Gaming, 64GB of G.Skill Ripjaws V DDR4-3600 running at 3200MHz because the higher speed was unstable, a Zotac RTX 3080 Ti, a Samsung 970 EVO Plus 1TB boot drive, a Kingston NV2 2TB game drive, a 10TB Toshiba hard drive, and a Corsair HX1000i power supply that is more than ten years old. Could the PSU, GPU power cables, motherboard, memory, or another component be responsible for this partial system freeze?
5 Answers
Since the problem appears during GPU-heavy workloads, test the graphics card independently. Run a repeatable GPU benchmark, try a different driver version with a clean installation, and temporarily lower the GPU power limit or clock. If lowering power makes the freezes disappear, that points toward the GPU, its power delivery, or the PSU rather than Windows itself. Check the GPU power connectors for heat damage or looseness.
A lack of a dump is expected with a hard lock: Windows may never get the chance to write one. Enable Kernel or Complete memory dumps if possible, but focus first on isolation. Run the GPU with reduced power, test with only the boot drive connected, use default BIOS settings, and swap in a known-good PSU. Change one thing at a time so you can identify which component affects the problem.
Test the memory with the system at completely stock settings. Disable DOCP/XMP, run a full memory test from a bootable tool, and test one DIMM at a time in the recommended motherboard slot. Memory instability can hard-freeze a system without creating a minidump, particularly when the CPU's memory controller is being stressed. Also update the motherboard BIOS and chipset drivers before retesting.
The save operation hanging could also indicate storage or controller trouble. Check the health and firmware of both NVMe drives, inspect SMART data, and temporarily disconnect the secondary drives one at a time. Look for storage reset or timeout events in Event Viewer, and avoid relying only on a quick benchmark because intermittent drive or controller failures may not show up every time.
The PSU is a serious suspect given its age, especially with a 3080 Ti's transient power spikes. A 1000W unit should be sufficient on paper, but an old PSU can become unreliable without producing a clear error. Try a known-good modern PSU and use separate PCIe power cables for the GPU rather than daisy-chaining connectors. Also make sure every connector is fully seated at both ends if the PSU is modular.
I have been monitoring temperatures and they look normal. Reseating the GPU power connectors seemed to help temporarily, so I am trying a replacement PCIe cable and will test again with separate cables.